Question - Depth anything model size

[SimonR99]

In your example, you used depth anything large but in your paper, you mentioned the small version. Did you see some improvement when using DAN-L ?

[chouss911]

In our experiment, we found that the size of the depth map model (s and l) did not have a significant impact on the final spatial clip effect. This is largely due to the fact that even the S model still provides sufficient depth information.